Eur Radiol DOI 10.1007/s00330-014-3192-z
COMPUTED TOMOGRAPHY
Can single-phase dual-energy CT reliably identify adrenal adenomas? A. Helck & N. Hummel & F. G. Meinel & T. Johnson & K. Nikolaou & A. Graser
Received: 13 December 2013 / Revised: 10 March 2014 / Accepted: 15 April 2014 # European Society of Radiology 2014
Abstract Purpose To evaluate whether single-phase dual-energy-CTbased attenuation measurements can reliably differentiate lipid-rich adrenal adenomas from malignant adrenal lesions. Materials and methods We retrospectively identified 51 patients with adrenal masses who had undergone contrastenhanced dual-energy-CT (140/100 or 140/80 kVp). Virtual non-contrast and colour-coded iodine images were generated, allowing for measurement of pre- and post-contrast density on a single-phase acquisition. Adrenal adenoma was diagnosed if density on virtual non-contrast images was ≤10 HU. Clinical follow-up, true non-contrast CT, PET/CT, in- and opposed-phase MRI, and histopathology served as the standard of reference. Results Based on the standard of reference, 46/57 (80.7 %) adrenal masses were characterised as adenomas or other benign lesions; 9 malignant lesions were detected. Based on a cutoff value of 10 HU, virtual non-contrast images allowed for correct identification of adrenal adenomas in 33 of 46 (71 %), whereas 13/46 (28 %) adrenal adenomas were lipid poor with a density ≥10 HU. Based on the threshold of 10 HU on the virtual non-contrast images, the sensitivity, specificity, and accuracy for detection of benign adrenal lesions was 73 %, 100 %, and 81 % respectively. Conclusion Virtual non-contrast images derived from dualenergy-CT allow for accurate characterisation of lipid-rich adrenal adenomas and can help to avoid additional follow-up imaging.
A. Helck and N. Hummel contributed equally to this work. A. Helck (*) : N. Hummel : F. G. Meinel : T. Johnson : K. Nikolaou : A. Graser Institute for Clinical Radiology, University of Munich, Grosshadern Campus, Marchioninistr. 15, 81377 Munich, Germany e-mail: [email protected]
Key Points • Adrenal adenomas are a common lesion of the adrenal glands. • Differentiation of benign adrenal adenomas from malignant adrenal lesions is important. • Dual-energy based virtual non-contrast images help to evaluate patients with adrenal adenomas. Keywords Dual-energy CT . Virtual non-contrast images . Adrenal lesion . Adrenal adenoma . Reduction of dose exposure
Introduction Adrenal masses are common pathologies with an incidence between 4.4 % and 9.0 % [1–5] on abdominal CTs. Since cross-sectional abdominal imaging is increasingly used, incidental adrenal masses are discovered more frequently. The majority of adrenal masses are benign adenomas. However, several malignant tumours metastasise to the adrenal glands and up to 2.5 % of adrenal incidentalomas turn out to be metastases [6, 7]. Hence, the differentiation between benign and malignant adrenal masses is of great importance, especially in patients with known malignancy. In principle, the diagnosis of a benign adenoma can easily be established by detection of intracellular fat on non-contrast CT images as about 70 % of adrenal adenomas contain a large amount of intracellular fat (mainly cholesterol and fatty acids) [8]. The diagnosis of this sort of adenoma can be assumed with a sensitivity of 70 % and specificity of 98 % when measurements show Hounsfield numbers lower than 10 on non-contrast CT. This has been proven by various studies showing that lipid-rich adenomas can be accurately differentiated from malignant lesions on
Eur Radiol
non-contrast CT images because of their low density [9–11]. However, in clinical routine, few abdominal CT examinations include a non-contrast acquisition, and contrast enhancement frequently masks fatty contents in lipid-rich adenomas, thereby making density measurements unreliable. In this setting, further imaging is required for accurate characterisation of unclear adrenal masses: either a non-contrast CT examination or a delayed phase examination showing fast washout of contrast agent has to be performed [6]. However, both procedures are time consuming and cause additional radiation exposure. Another option for adrenal mass characterisation is MRI chemical shift imaging using an in and opposed phase [8, 12]. This type of additional imaging is time consuming and associated with significant additional costs. An ideal solution for this dilemma is dual-energy CT. The technology has shown great potential in numerous clinical indications in the abdomen [13]. Since it enables subtraction of iodine from a contrast-enhanced CT examination by means of three material decomposition techniques, it provides the same diagnostic information as a biphasic CT examination containing a noncontrast and a contrast-enhanced phase [14, 15]. This information could be used to characterise unclear adrenal masses on a single-phase examination [12, 16]. The purpose of this study was to assess the accuracy of DECT in the differentiation of lipid-rich adrenal adenomas from malignant adrenal masses using single-phase dual-energy contrast-enhanced abdominal CT.
Materials and methods Patient population Our in-house patient database (Syngo Imaging/Syngo RIS, Siemens AG Healthcare, Forchheim, Germany) was accessed using specific search terms ("dual energy" AND “adrenal” ~ OR “adrenal adenoma”~ OR “adrenal lesion”~OR “incidentaloma”~OR “adrenal metastasis”) in order to identify patients who (1) had undergone abdominal DECT and (2) had reported abnormalities of the adrenal glands. Thus, we retrospectively identified all patients with adrenal masses of at least 1 cm who had undergone single-phase contrast-enhanced DECT examinations of the abdomen and included these individuals in the study. IRB approval was waived. CT protocol CT imaging was performed on one of two dual-source multidetector row CT systems (Somatom Definition Dual Source, n=x or Somatom Definition Flash, n=y; Siemens Healthcare, Forchheim, Germany). Each of these systems consists of two
X-ray tubes and their corresponding detectors mounted in an approximately 90-degree angle. Patients were positioned on the CT table in the supine position. The Somatom Definition (first-generation system) had a field of view of 26 cm (FOV) on its smaller detector (26 cm), whereas the Somatom Definition Flash (second-generation system) provides a FOV of 33 cm on the smaller detector. The larger detectors have an identical size and cover 50 cm; 70 s after intravenous injection of a nonionic contrast agent (1.35 ml/kg patient body weight; Ultravist 370, BayerSchering Diagnostics, Berlin, Germany), a dual-energy examination of the abdomen was acquired operating the “A” tube at 140 kVp (140 kVp with tin filtering, Sn140 kVp for the Flash CT) and the “B” tube at 80 kVp (100 kVp for the Flash CT). For both tubes, an automated attenuation-based tube current modulation (CareDOSE 4D, Siemens, Forchheim, Germany) was used. All examinations were acquired at a pitch of 0.55 to 0.6; the gantry rotation speed was 0.5 s [17, 18]. The dose length products (DLPs) were recorded from the patient protocol and were used for calculation of estimates of individual effective radiation doses in mSv using an organ-based standard conversion factor for abdominal CT (0.015 mSv/mGy*cm).
Dual-energy post processing and image reconstruction From one DECT examination, three different sets of images are created: one low- and one high-kVp data set, and weighted average images, which are based on data from both detectors, using 70 % information from the high kV and 30 % from the low kV examination (Definition DS) and 50 % from each kV examination (Definition Flash). These images resemble the Table 1 The imaging protocol was adapted to the type of CT CT parameters for the two dual-source CTs Definition DS (tubes A/B)
Definition Flash (tubes A/B)
Slices Angular offset (°) FoV (cm) Spectra (kVp) Filter (mm)
2×64 90 50/26 140/80 3 A1, 0.9 Ti
Current time product (mAs) Modulation
96/404
2×128 95 50/33 100/Sn140 3 A1, 0.9, Ti, 0.4 Sn (B tube only) 204/140
CareDose4D (x, y, z-axis) 14×1.2 0.55 0.5
CareDose 4D (x, y, z-axis) 32×0.6 0.6 0.5
Collimation (mm) Pitch Rotation time (s)
Definition DS indicates first-generation dual-source CT; Definition FLASH, second-generation dual-source CT
Eur Radiol
image quality of a standard 120-kVp CT examination. For post processing, a dedicated dual-energy workstation (syngo Multi-Modality Workplace, Somaris Version CT2008G; Siemens Healthcare) was used. An application called “Liver VNC” was utilised to create images from which iodine had been subtracted [15, 17]. This calculation is based on a so-called three-material decomposition analysis [19]. The 3-mm-thick virtual noncontrast images were reconstructed from the raw data using a 2-mm reconstruction increment to allow for exact three-material decompositions at limited image noise. Analysis of adrenal lesions For assessment of lesion size, a volumetric analysis using the “volume” software (syngo MMWP, Siemens Healthcare, Erlangen/Germany) of each adrenal lesion was performed. The density of all adrenal lesions was measured on both Table 2 Patient characteristics with underlying disease and final diagnoses
Characterisation of adrenal masses as benign or malignant For the assessment of lesion aetiology clinical follow-up (n=12), true non-contrast CT (n=26), PET/CT (n=5), inand opposed-phase MRI (n=5), and histopathology (n=9) were used. A benign lesion was diagnosed when there was lack of size progression for at least 6 months. In ten patients without medical history of malignancy, no follow-up imaging was performed. In these cases a clinical follow-up of at least 18 months was considered diagnostic for absence of malignancy [3, 8, 20, 21].
n
Underlying disease
Adenoma (n=45)
Other adrenal lesions (n=3)
Metastasis (n=8)
Malignant 10 5 4 3 2
Renal cell carcinoma GIST Pancreatic ductal adenocarcinoma Colorectal cancer Oropharyngeal squamous cell carcinoma
7 5 2 3 1
-
3 2 1
Acute lymphocytic leukaemia Adenocarcinoma of the lung Chondrosarcoma Gastric adenocarcinoma Invasive ductal carcinoma (breast) Ovarian carcinoma Pheochromocytoma Seminoma
1 1 1 1 1
Pheochromocytoma -
1* 1 -
Abdominal aortic aneurysm Unknown lesion of the renal bed Coronary artery disease (CAD) Hormone-active adrenal adenoma Lower back pain Anaemia of unknown origin Arterial hypertension Atrial septal defect Barrett’s oesophagus Benign prostatic hyperplasia
5 4 2 2 2 1 1 1 1 1
Lipoangioma -
-
Heart transplantation Peripheral arterial occlusive disease Pulmonary embolism Unknown lymphadenopathy
1 1 1
Myelolipoma -
-
1 1 1 1 1 1 1 1 Benign 6 4 2 2 2 1 1 1 1 1
* Infiltration
contrast-enhanced and virtual non-contrast images. The mean HU was calculated by performing three repetitive measurements; the respective ROI was placed centrally in the lesion and was adapted to lesion size to avoid a partial volume effect.
1 1 1 1
Eur Radiol
Statistical analysis Sensitivity, specificity, and accuracy of dual-energy virtual non-contrast images for the characterisation of incidental adrenal masses were calculated using chi-square tables of contingency.
Results In the study period, n=457 abdominal dual-energy CT examinations were performed at our department. In 51 of these patients, adrenal incidentalomas were detected and these individuals formed the study population. DECT examinations were performed between April 2007 and September 2012. The patient population consisted of 28 men and 23 women (mean age, 70±13 years; range, 24−85 years); 53 % of patients (n=27) had adrenal masses on the left side, 35 % (n=18) on the right, and in 12 % (n=6) both sides were affected. In total, 57 adrenal masses were identified. Tables 1 and 2 describes patient characteristics in detail and gives more information about the final diagnosis. Based on the reference standard, 48 adrenal masses turned out to be benign lesions and in 9 patients malignant lesions were diagnosed. All lesions with HU
COMPUTED TOMOGRAPHY
Can single-phase dual-energy CT reliably identify adrenal adenomas? A. Helck & N. Hummel & F. G. Meinel & T. Johnson & K. Nikolaou & A. Graser
Received: 13 December 2013 / Revised: 10 March 2014 / Accepted: 15 April 2014 # European Society of Radiology 2014
Abstract Purpose To evaluate whether single-phase dual-energy-CTbased attenuation measurements can reliably differentiate lipid-rich adrenal adenomas from malignant adrenal lesions. Materials and methods We retrospectively identified 51 patients with adrenal masses who had undergone contrastenhanced dual-energy-CT (140/100 or 140/80 kVp). Virtual non-contrast and colour-coded iodine images were generated, allowing for measurement of pre- and post-contrast density on a single-phase acquisition. Adrenal adenoma was diagnosed if density on virtual non-contrast images was ≤10 HU. Clinical follow-up, true non-contrast CT, PET/CT, in- and opposed-phase MRI, and histopathology served as the standard of reference. Results Based on the standard of reference, 46/57 (80.7 %) adrenal masses were characterised as adenomas or other benign lesions; 9 malignant lesions were detected. Based on a cutoff value of 10 HU, virtual non-contrast images allowed for correct identification of adrenal adenomas in 33 of 46 (71 %), whereas 13/46 (28 %) adrenal adenomas were lipid poor with a density ≥10 HU. Based on the threshold of 10 HU on the virtual non-contrast images, the sensitivity, specificity, and accuracy for detection of benign adrenal lesions was 73 %, 100 %, and 81 % respectively. Conclusion Virtual non-contrast images derived from dualenergy-CT allow for accurate characterisation of lipid-rich adrenal adenomas and can help to avoid additional follow-up imaging.
A. Helck and N. Hummel contributed equally to this work. A. Helck (*) : N. Hummel : F. G. Meinel : T. Johnson : K. Nikolaou : A. Graser Institute for Clinical Radiology, University of Munich, Grosshadern Campus, Marchioninistr. 15, 81377 Munich, Germany e-mail: [email protected]
Key Points • Adrenal adenomas are a common lesion of the adrenal glands. • Differentiation of benign adrenal adenomas from malignant adrenal lesions is important. • Dual-energy based virtual non-contrast images help to evaluate patients with adrenal adenomas. Keywords Dual-energy CT . Virtual non-contrast images . Adrenal lesion . Adrenal adenoma . Reduction of dose exposure
Introduction Adrenal masses are common pathologies with an incidence between 4.4 % and 9.0 % [1–5] on abdominal CTs. Since cross-sectional abdominal imaging is increasingly used, incidental adrenal masses are discovered more frequently. The majority of adrenal masses are benign adenomas. However, several malignant tumours metastasise to the adrenal glands and up to 2.5 % of adrenal incidentalomas turn out to be metastases [6, 7]. Hence, the differentiation between benign and malignant adrenal masses is of great importance, especially in patients with known malignancy. In principle, the diagnosis of a benign adenoma can easily be established by detection of intracellular fat on non-contrast CT images as about 70 % of adrenal adenomas contain a large amount of intracellular fat (mainly cholesterol and fatty acids) [8]. The diagnosis of this sort of adenoma can be assumed with a sensitivity of 70 % and specificity of 98 % when measurements show Hounsfield numbers lower than 10 on non-contrast CT. This has been proven by various studies showing that lipid-rich adenomas can be accurately differentiated from malignant lesions on
Eur Radiol
non-contrast CT images because of their low density [9–11]. However, in clinical routine, few abdominal CT examinations include a non-contrast acquisition, and contrast enhancement frequently masks fatty contents in lipid-rich adenomas, thereby making density measurements unreliable. In this setting, further imaging is required for accurate characterisation of unclear adrenal masses: either a non-contrast CT examination or a delayed phase examination showing fast washout of contrast agent has to be performed [6]. However, both procedures are time consuming and cause additional radiation exposure. Another option for adrenal mass characterisation is MRI chemical shift imaging using an in and opposed phase [8, 12]. This type of additional imaging is time consuming and associated with significant additional costs. An ideal solution for this dilemma is dual-energy CT. The technology has shown great potential in numerous clinical indications in the abdomen [13]. Since it enables subtraction of iodine from a contrast-enhanced CT examination by means of three material decomposition techniques, it provides the same diagnostic information as a biphasic CT examination containing a noncontrast and a contrast-enhanced phase [14, 15]. This information could be used to characterise unclear adrenal masses on a single-phase examination [12, 16]. The purpose of this study was to assess the accuracy of DECT in the differentiation of lipid-rich adrenal adenomas from malignant adrenal masses using single-phase dual-energy contrast-enhanced abdominal CT.
Materials and methods Patient population Our in-house patient database (Syngo Imaging/Syngo RIS, Siemens AG Healthcare, Forchheim, Germany) was accessed using specific search terms ("dual energy" AND “adrenal” ~ OR “adrenal adenoma”~ OR “adrenal lesion”~OR “incidentaloma”~OR “adrenal metastasis”) in order to identify patients who (1) had undergone abdominal DECT and (2) had reported abnormalities of the adrenal glands. Thus, we retrospectively identified all patients with adrenal masses of at least 1 cm who had undergone single-phase contrast-enhanced DECT examinations of the abdomen and included these individuals in the study. IRB approval was waived. CT protocol CT imaging was performed on one of two dual-source multidetector row CT systems (Somatom Definition Dual Source, n=x or Somatom Definition Flash, n=y; Siemens Healthcare, Forchheim, Germany). Each of these systems consists of two
X-ray tubes and their corresponding detectors mounted in an approximately 90-degree angle. Patients were positioned on the CT table in the supine position. The Somatom Definition (first-generation system) had a field of view of 26 cm (FOV) on its smaller detector (26 cm), whereas the Somatom Definition Flash (second-generation system) provides a FOV of 33 cm on the smaller detector. The larger detectors have an identical size and cover 50 cm; 70 s after intravenous injection of a nonionic contrast agent (1.35 ml/kg patient body weight; Ultravist 370, BayerSchering Diagnostics, Berlin, Germany), a dual-energy examination of the abdomen was acquired operating the “A” tube at 140 kVp (140 kVp with tin filtering, Sn140 kVp for the Flash CT) and the “B” tube at 80 kVp (100 kVp for the Flash CT). For both tubes, an automated attenuation-based tube current modulation (CareDOSE 4D, Siemens, Forchheim, Germany) was used. All examinations were acquired at a pitch of 0.55 to 0.6; the gantry rotation speed was 0.5 s [17, 18]. The dose length products (DLPs) were recorded from the patient protocol and were used for calculation of estimates of individual effective radiation doses in mSv using an organ-based standard conversion factor for abdominal CT (0.015 mSv/mGy*cm).
Dual-energy post processing and image reconstruction From one DECT examination, three different sets of images are created: one low- and one high-kVp data set, and weighted average images, which are based on data from both detectors, using 70 % information from the high kV and 30 % from the low kV examination (Definition DS) and 50 % from each kV examination (Definition Flash). These images resemble the Table 1 The imaging protocol was adapted to the type of CT CT parameters for the two dual-source CTs Definition DS (tubes A/B)
Definition Flash (tubes A/B)
Slices Angular offset (°) FoV (cm) Spectra (kVp) Filter (mm)
2×64 90 50/26 140/80 3 A1, 0.9 Ti
Current time product (mAs) Modulation
96/404
2×128 95 50/33 100/Sn140 3 A1, 0.9, Ti, 0.4 Sn (B tube only) 204/140
CareDose4D (x, y, z-axis) 14×1.2 0.55 0.5
CareDose 4D (x, y, z-axis) 32×0.6 0.6 0.5
Collimation (mm) Pitch Rotation time (s)
Definition DS indicates first-generation dual-source CT; Definition FLASH, second-generation dual-source CT
Eur Radiol
image quality of a standard 120-kVp CT examination. For post processing, a dedicated dual-energy workstation (syngo Multi-Modality Workplace, Somaris Version CT2008G; Siemens Healthcare) was used. An application called “Liver VNC” was utilised to create images from which iodine had been subtracted [15, 17]. This calculation is based on a so-called three-material decomposition analysis [19]. The 3-mm-thick virtual noncontrast images were reconstructed from the raw data using a 2-mm reconstruction increment to allow for exact three-material decompositions at limited image noise. Analysis of adrenal lesions For assessment of lesion size, a volumetric analysis using the “volume” software (syngo MMWP, Siemens Healthcare, Erlangen/Germany) of each adrenal lesion was performed. The density of all adrenal lesions was measured on both Table 2 Patient characteristics with underlying disease and final diagnoses
Characterisation of adrenal masses as benign or malignant For the assessment of lesion aetiology clinical follow-up (n=12), true non-contrast CT (n=26), PET/CT (n=5), inand opposed-phase MRI (n=5), and histopathology (n=9) were used. A benign lesion was diagnosed when there was lack of size progression for at least 6 months. In ten patients without medical history of malignancy, no follow-up imaging was performed. In these cases a clinical follow-up of at least 18 months was considered diagnostic for absence of malignancy [3, 8, 20, 21].
n
Underlying disease
Adenoma (n=45)
Other adrenal lesions (n=3)
Metastasis (n=8)
Malignant 10 5 4 3 2
Renal cell carcinoma GIST Pancreatic ductal adenocarcinoma Colorectal cancer Oropharyngeal squamous cell carcinoma
7 5 2 3 1
-
3 2 1
Acute lymphocytic leukaemia Adenocarcinoma of the lung Chondrosarcoma Gastric adenocarcinoma Invasive ductal carcinoma (breast) Ovarian carcinoma Pheochromocytoma Seminoma
1 1 1 1 1
Pheochromocytoma -
1* 1 -
Abdominal aortic aneurysm Unknown lesion of the renal bed Coronary artery disease (CAD) Hormone-active adrenal adenoma Lower back pain Anaemia of unknown origin Arterial hypertension Atrial septal defect Barrett’s oesophagus Benign prostatic hyperplasia
5 4 2 2 2 1 1 1 1 1
Lipoangioma -
-
Heart transplantation Peripheral arterial occlusive disease Pulmonary embolism Unknown lymphadenopathy
1 1 1
Myelolipoma -
-
1 1 1 1 1 1 1 1 Benign 6 4 2 2 2 1 1 1 1 1
* Infiltration
contrast-enhanced and virtual non-contrast images. The mean HU was calculated by performing three repetitive measurements; the respective ROI was placed centrally in the lesion and was adapted to lesion size to avoid a partial volume effect.
1 1 1 1
Eur Radiol
Statistical analysis Sensitivity, specificity, and accuracy of dual-energy virtual non-contrast images for the characterisation of incidental adrenal masses were calculated using chi-square tables of contingency.
Results In the study period, n=457 abdominal dual-energy CT examinations were performed at our department. In 51 of these patients, adrenal incidentalomas were detected and these individuals formed the study population. DECT examinations were performed between April 2007 and September 2012. The patient population consisted of 28 men and 23 women (mean age, 70±13 years; range, 24−85 years); 53 % of patients (n=27) had adrenal masses on the left side, 35 % (n=18) on the right, and in 12 % (n=6) both sides were affected. In total, 57 adrenal masses were identified. Tables 1 and 2 describes patient characteristics in detail and gives more information about the final diagnosis. Based on the reference standard, 48 adrenal masses turned out to be benign lesions and in 9 patients malignant lesions were diagnosed. All lesions with HU
